Onyinyechi Nneji

May 24, 2021

7 min read

Take your digital marketing campaigns to the next level using attribution modelling

Early this year, I worked on a project for a client. The goal of the project was to get people to sign-up for a new product the company had just launched. The client engaged a designer, a developer and me, a growth marketer to ensure that at least 50,000 people signed up in 1 month. Given the advertising budget and how useful the product was, 50,000 signups hardly felt unattainable. So I immediately got to work. Having no product manager, I set up a meeting with the designer and the developer to discuss our thought process, I wanted to make sure we were all aligned on the next steps.

Following this, we created a landing page with a very simple form to gather information and I created social media and AdWords collaterals. One of the objectives of the campaign was also product awareness, so, the client didn’t just want people signing up for the free version of the product, the product being talked about by people who fit the target audience was also a priority. For this objective, I recommended the production of video content just for this campaign, demonstrating the product’s impact on the target audience. I set up a multi-channel campaign. After a week, the landing page had been created, the video had been shot and the campaign was running. After 4 weeks, I submitted a report to the client, all metrics had been met and surpassed.

About a month later, I got a call from the client, the product launch had gone so well, they wanted the team to work together to put more money behind advertising the product and in their words, they wanted to redo the campaign on a way bigger scale. They were considering about 3x the budget spent on the first campaign. They wanted more videos, more social media ads, more articles, they wanted to replicate the campaign exactly as it had been executed but I had a better idea.

When I set up digital marketing for a client, I do it with growth in mind. In this situation, what this meant was, because I knew I was using a lot of channels and a huge budget, I had prioritized choosing the right attribution model. It was important to me not just that I deliver the objectives of 50,000 sign-ups in a month but to understand user behaviour as I did that. I wanted to know which channels worked best and delivered the best results. What Ads worked best, what blogs brought in the most sign-ups, did the video influence user behaviour and how did people navigate through the landing page.

What is attribution modelling?

The term attribution modelling describes the strategies marketers use to measure and assign credit to marketing touchpoints including those attributed to a product search, a purchase, and every action in between. Using attribution models helps marketers better understand which parts of their marketing effort are driving the most leads to that part of the sales funnel.

In this situation, I was running this campaign across multiple platforms and it was important to understand which mix of PPC keywords, display ads, landing pages, SEO, Social Media Ads were generating leads.

For example, a customer could have seen just one social media ad and immediately signed while another customer watched a video, saw the social media ad but finally converted when she saw an ad banner on one of her favourite blogs.

Why do you need Attribution Modelling?

As a growth marketer, If you only give credit to the last touchpoint before a conversion, you’re missing out on all the other channels that influenced your customer’s choice to convert. It’s important to understand how and when your marketing efforts create conversion.

Attribution modelling can provide vital data about your marketing efforts, including what paid search campaigns work, what campaigns don’t, and where you should be allocating your money for optimal conversion rates at important milestones in the sales process. It allows you to work smarter not harder, and harness the power of return on investment (ROI) data.

Types of Attribution Models

There’s a large variety of attribution models out there, from AI-based software, Facebook’s own attribution tool, or the standard Google Analytics attribution models. However, I’ll go through the most popular ones.

  • Last interaction attribution model
  • Linear attribution model
  • Time decay attribution model
  • Position Based attribution model

Last Non-direct click:

Last non-direct click attribution models give 100% of the credit to your customers’ last non-direct touchpoint. Non-direct traffic is all traffic that’s been guided to your website from another source. Your email campaigns, Twitter posts, influencer marketing campaigns, etc. should all be accompanied by a corresponding UTM.

For example, if a customer read a blog post about the new product and then a few days later decides to visit the landing page to signup, the credit will be given to the blog post.

In analytics, in AdWords and in a number of different platforms the credit of that conversion will not go to direct, even though the customer directly visited the landing page. If there is less than 30 days between the last non-direct click or visit before that direct visit where the conversion was driven, then the conversion credit goes to the last non-direct click.

Last Click attribution model:

This model gives all of the credit to the last interaction your business had with a customer before they convert i.e. register, download, make a purchase etc. I think this is is actually a slightly better way of looking at performance compared to the last non-direct because it is more direct.

Last click attribution also is really useful when you start talking about linking the offline world to online because the last interaction doesn’t necessarily have to be digital. In this project, some people had walked into the client’s office to pay for the product.

Conversion won’t always occur digitally therefore it is really important to have the right model for your business.

Linear Attribution Model:

This model involves distributing the value equally amongst all of the touchpoints in a conversion path or customer journey. If there were 3 touchpoints in the consumer journey, each of those points would get 33.33% of the overall credit for that sale. This is the equivalent of 1000 naira conversion being given 333 naira for every single interaction. Therefore if every visit was a different channel, you would split that back through the journey equally.

This model benefits all channels because it will actually identify what channels contributed to the conversion. However, there’s quite a large flaw with this model because it does it equally and doesn’t show what channels actually impacted that user.

Therefore, you can’t really use this to understand why someone interacted with that activity and did it actually make a difference to their decision making.

Time Decay Attribution Model:

This model is similar to the “linear attribution” model in that each of the touchpoints is ascribed some credit for a conversion, however, the most recent touchpoints are given more of the pie, and the least recent interactions get a smaller piece.

This model has a lot of value to many organizations. The logic of when someone did something is super important and it’s actually not utilized that much at the moment.The last interaction has the most value in this model but what this is also saying is we want to take into consideration how long back through that journey that previous interaction happened.

Position-Based Attribution Model

This model was popularized by Google Analytics being the most accurate. This model divides up the credit for a conversion between a customer’s first interaction with your brand and the moment they convert to a lead, with each receiving 40%. The remaining 20% of the credit is spread out between all other interactions that happened between the first interaction and the sale.

This is typically what we see when we calculate and speak to people about where they think the value should be distributed. The first time that you engage with a brand, you’re going to be able to say, this is how I found out about this company. And the last thing, the last interaction is where you’ll say, this is where I converted and therefore everything in the middle has a significantly lower value.

This model has a lot of benefits because it does align very well with how consumers actually behave. The position-based model gives us an understanding of the challenges of acquisition and the challenges of conversion.

In conclusion, deciding on what attribution model to use depends on how complicated your average consumer journey is. If you’re a large company you should opt for a comprehensive model but if you are a relatively small organization, then you probably don’t need to invest in a highly complex attribution model.

In my case, setting up an effective attribution model meant that when the client came back, I didn’t have to just set up the same campaign on a bigger scale but I could set up a smarter and more effective marketing campaign, based on results from the first campaign. My campaign was smarter and got the desired result while saving cost for the client.

Hi, Onyinyechi Nneji here! I am a user-focused growth and digital marketing professional. I create strategies with one goal in mind, driving users to perform actions that align with business goals. If you’d like to reach out, follow me on Twitter or send me an email.